Online data storage/hosting

[ciggwin]

Hello again

I am in the market for a new data server. We currently have about 350 GB of data that I would need to migrate over. Right now it's spread out between a server and a NAS device. I kind of talked about this in the backup thread I made, but now management has me on another wild goose chase INTO THE CLOUD!!!!

They want to know why we can't have our data online somewhere... like a giant Google Docs for all of our data, from which users will work. Not only have I never heard of this before, I have to quote it out and present why it isn't feasible. Has anyone even HEARD of this? I am not talking Dropbox type stuff, I am talking about having NO data server at the office and having everything on a remote file system somewhere on the Internet. And no I am not talking about backup, I am talking about data hosting.

"That way, everyone has access 24/7!!!! I mean what if the VPN goes down????" - Management

Not only is this idea... completely retarded, I'm not allowed to purchase anything until I give price quotes and an entire report on why online data hosting is or is not a viable data storage solution. They also think there is some sort of cost savings as if you host your data online, backup is going to be built-in and we won't have to pay for a backup system/online backup.

I can come up with reasons why this idea is not a good one, but I've actually done some searching online and I don't see anything that resembles what they are looking for. If you told me that a company does not have a data server, I would probably give a look like you will find in one of those WTF?!? photos with the dog or cat.

Anybody have experience with this? I realize this is not hardware related and that it is basically the opposite of [H]ard So my apologies.

 

[pyrodex]

Hello again

I am in the market for a new data server. We currently have about 350 GB of data that I would need to migrate over. Right now it's spread out between a server and a NAS device. I kind of talked about this in the backup thread I made, but now management has me on another wild goose chase INTO THE CLOUD!!!!

They want to know why we can't have our data online somewhere... like a giant Google Docs for all of our data, from which users will work. Not only have I never heard of this before, I have to quote it out and present why it isn't feasible. Has anyone even HEARD of this? I am not talking Dropbox type stuff, I am talking about having NO data server at the office and having everything on a remote file system somewhere on the Internet. And no I am not talking about backup, I am talking about data hosting.

"That way, everyone has access 24/7!!!! I mean what if the VPN goes down????" - Management

Not only is this idea... completely retarded, I'm not allowed to purchase anything until I give price quotes and an entire report on why online data hosting is or is not a viable data storage solution. They also think there is some sort of cost savings as if you host your data online, backup is going to be built-in and we won't have to pay for a backup system/online backup.

I can come up with reasons why this idea is not a good one, but I've actually done some searching online and I don't see anything that resembles what they are looking for. If you told me that a company does not have a data server, I would probably give a look like you will find in one of those WTF?!? photos with the dog or cat.

Anybody have experience with this? I realize this is not hardware related and that it is basically the opposite of [H]ard So my apologies.

I personally do this with my side business using Amazon Storage, it is cheap and you pay per GB stored. There are various clients out there that mount the Amazon storage as a regular drive share in Windows/Mac and then users can work like normal.

 

[ciggwin]

I personally do this with my side business using Amazon Storage, it is cheap and you pay per GB stored. There are various clients out there that mount the Amazon storage as a regular drive share in Windows/Mac and then users can work like normal.

But is this really a solution that is appropriate for 50+ users?

I did find storsimple but for the price I found on that they're going to cringe. Good for me, I guess.

 

[pyrodex]

But is this really a solution that is appropriate for 50+ users?

I did find storsimple but for the price I found on that they're going to cringe. Good for me, I guess.

We do it with about 10 people but it depends on your data. Don't forget with some cloud solutions you may lose the ability to grant/remove permissions so you will need an expensive enterprise solution .

 

[danman]

Well the definition of 'the cloud' is just someone else's datacenter. You either have the data on premise under your total control, co-located with other servers off premise, or located on someone else's servers.

 

[vraa]

I can't imagine any company running 100% offsite
Even with multiple wan...

 

[war9200]

Hi,

Its more common than you may think, its called Cross Site.

DC1 -> DC2
DC2 -> DC1

 

[Chimel]

Give it a go, Amazon S3 is the IT admin and accountant best friend, cheap too: $35 to copy your 350 GB, $50/month of storage fee, and another cheap monthly transfer fee.

With your own internal solution, you need to purchase, maintain, renew a server, a RAID system with failing drives, a backup system, network connections, etc. It's a deprecable hardware and software cost, plus monthly maintenance costs, plus extraordinary costs, a nightmare to manage for accounting.

S3 was initially created for developers, I don't know how usable it is for corporate documents. For instance if you can create different sets of permissions for different users within a S3 "bucket." With the free basic offer, it's easy to try for yourself at no cost but your own time, and get help from the S3 forum.

 

[drescherjm]

With your own internal solution, you need to purchase, maintain, renew a server, a RAID system with failing drives, a backup system, network connections, etc. It's a deprecable hardware and software cost, plus monthly maintenance costs, plus extraordinary costs, a nightmare to manage for accounting.

Depends on what your costs are and how much maintenance you do. For me a typical $1200 US 5 to 7TB linux raid 5/6 server takes about 10 total minutes a month in maintenance.

 

[ciggwin]

I can't imagine any company running 100% offsite
Even with multiple wan...

I can't either. They want to essentially eliminate the entire infrastructure. Not only does that eliminate the need for a full time IT person (me) but it's just not something I want to get into. I didn't pursue the IT profession to outsource everything so I don't have a job. I want to do stuff that relates to my field.

Hi,

Its more common than you may think, its called Cross Site.

DC1 -> DC2
DC2 -> DC1

If we had a DC, yes, I could see this but we are a one stop shop. Think mom & pop 50 employee company. Very basic.

Give it a go, Amazon S3 is the IT admin and accountant best friend, cheap too: $35 to copy your 350 GB, $50/month of storage fee, and another cheap monthly transfer fee.

With your own internal solution, you need to purchase, maintain, renew a server, a RAID system with failing drives, a backup system, network connections, etc. It's a deprecable hardware and software cost, plus monthly maintenance costs, plus extraordinary costs, a nightmare to manage for accounting.

S3 was initially created for developers, I don't know how usable it is for corporate documents. For instance if you can create different sets of permissions for different users within a S3 "bucket." With the free basic offer, it's easy to try for yourself at no cost but your own time, and get help from the S3 forum.

I read up on S3 and it seems like it would work but I need to look into permissions. I'd have to test it out with my IT folder before moving everyone to it. I am just being stubborn and I don't want it

Depends on what your costs are and how much maintenance you do. For me a typical $1200 US 5 to 7TB linux raid 5/6 server takes about 10 total minutes a month in maintenance.

Agree - the data server we have is 5 years old and I haven't had to do anything to it since I started over a year ago. They're paying me to be here anyway - I don't really see an increased "cost" from general maintenance since I'm usually browsing [H] instead of having work to do. The real factor in the "argument" I have with management is going to be two-fold: price and why the buzzword of "cloud" is not necessarily the best option. Ever hear of the Electronic Communications Privacy Act of 1986?

 

[GeorgeHR]

One problem with "cloud" storage is what to do when someone makes a mistake.

So someone makes changes to a 6 year old file and realizes that the change was a mistake. How do you get the original back?

Make sure there is an archive/recovery facility.

 

[vraa]

Also make sure you aren't wikileaks
Otherwise amazon will fuck you over

Just to confirm - you have explained that by moving EVERYTHING off-site things will be super slow and if the host is down you are fucked?
Ideally you should use your local as your primary, remote as redundant / secondary. Or if you are big baller you do master-master.

 

[ciggwin]

Also make sure you aren't wikileaks
Otherwise amazon will fuck you over

Just to confirm - you have explained that by moving EVERYTHING off-site things will be super slow and if the host is down you are fucked?
Ideally you should use your local as your primary, remote as redundant / secondary. Or if you are big baller you do master-master.

"But they have 99.99% uptime. What do we have here at the office?" - manager

 

[drescherjm]

delete

 

[vraa]

"But they have 99.99% uptime. What do we have here at the office?" - manager

What about the rest of the links in the system, they are no where near 99.99% uptime

Explain a simple concept, take 5 processes, each with 99% uptime
Your total reliability is bounded by (.99)^(5) = .95
Let's pretend you have 10 99% uptime links in the way -- (.99)^(10) = .90!!!
Play it safe, keep a local copy + snapshots + live replication remote/offsite. This is a solution that scales for tiny companies to huge companies. I must be missing more Let me know what your managers say

 

[ghost6303]

its really hard to tell if "the cloud" is a good place for your data without any information about what the data actually is and how its used. if its 350 gigs of PDF and word documents, maybe it could work. if its 350 gigs of databases, videos that are constantly edited or something that is bandwidth intensive or latency sensitive, maybe its best to keep in house.

you also need to be able to trust the security at whatever company you choose. i am extraordinarily suspicious when sending my data off into the wild blue yonder (cloud).

 

[Chimel]

Depends on what your costs are and how much maintenance you do. For me a typical $1200 US 5 to 7TB linux raid 5/6 server takes about 10 total minutes a month in maintenance.

I mean more from the accounting point of view, it is a financial advantage that push enterprises to rent rather than own. Same as for cars. That's the main reason why the OP's managers are asking. That and no doubt also because they just came back from a seminar where the magic word "cloud" was pronounced!
Another frequently used argument the OP might encounter is that outsourcing allows IT employees to focus more on IT for the core business, not peripheral tasks like storage.
And I don't mean to recommend one way or the other, just that Amazon S3 professional usage is spreading and is worth investigating when you have only 350 GB to manage.

Explain a simple concept, take 5 processes, each with 99% uptime
Your total reliability is bounded by (.99)^(5) = .95

It's 99.99% uptime, not 99%, so for 5 processes it's still 99.95%, 99.90% for 10.
Unless you set up a cluster of servers just for the 350 GB, you'll have downtime every Patch Tuesday; Even if it takes only 10 minutes to reboot every 4 weeks, you're down to 99.90% uptime. Now the relevance of being online 24/7/365 is another debate...

As for the comments about security, privacy, IP and other risks, I'm pretty sure they are much more important with owned hardware than with a cloud, it would be interesting to ask if Amazon has some studies to share about this.

 

[vraa]

I simplified it greatly, but you'll have a lot more than 5 links to fail throughout the process
One of the points is do not rely on a single point, you have to focus on all the parts of the chain
Do you already have multiple wans? Different providers? Are they all pretty fast (or can they be bonded?) I I think these are more of the basic questions which THEN you can decide - can you focus on a "cloud" system as your primary

BTW - I dunno what your exact requirements are, but the uptime on my ZFSGuru box is that of when I INSTALLED IT. Windows is not your only option, especially if you are talking about REOCCURING costs. SaaS is a bitch like that. I know it seems like I'm taking the position against s3, I'm not, I'm just saying don't use it as a primary, use it as a backup/secondary master

 

[pjkenned]

One thought for OP:

If you have 50 client PCs and something goes severely awry, requiring a re-image of all of them, you basically are going to be transferring 350GB at what? 10MB/s max sustained download on your WAN link for an organization that size? You can do the math, but that is the big risk.

With that being said... on site primary storage w/ S3 or similar off site is not a terrible idea.

 

[Lazn_Work]

I did move to hosting for some of my data, and it is the slowest most problematic of my data now. However it is no longer my problem. When something goes wrong I give the user a 1800 number and remind them that it wasn't my choice to do this.

Also 99% uptime sure, but at what time of day, if that 1% of down time is when you are trying to use it (and not at night on the weekend)...

Remember even with modern high speed connections to the interwebs further away will be slower no matter how fast the server on the other end is. So imagine a few seconds extra to open each file, etc. that all adds up and everyone's computer will feel slower.

It has it's uses and benefits, but mostly as a 2nd backup in my mind, not as the main or only data point.